CENTRAL EUROPEAN BUSINESS REVIEW

RESEARCH PAPERS

VOLUME 3, NUMBER 3 , SEPTEMBER 2014

THE RISK OF VAMPIRE EFFECT IN ADVERTISEMENTS

USING CELEBRITY ENDORSEMENT
Kuvita, T., Karlek, M.

In the race to get the attention of the target audience, advertisers often use special attention-getting devices. This in
turn exposes them to a higher risk of creating a vampire effect when the core message about the brand or a product
is eaten up by such devices. The concept of a vampire effect in advertising appears to be under-researched in
the current literature. Therefore, this paper provides deeper insights into the vampire effect occurrences in printed
advertisements using celebrity endorsement. The paper is based on a qualitative study with an eye-tracking device
with 12 participants and on the following experiment with 60 university students. The research found that a signicantly
higher risk of creating a vampire effect exists when using an unrelated celebrity as an attention-getting device than
when using a related celebrity or no celebrity at all. Marketers are advised to use related celebrities if choosing to stick
to this attention-getting approach. However, the concept of relatedness should be pre-tested prior to launching an
advertising campaign.
JEL classication: M370
Keywords: vampire effect; celebrity endorsement; attention-getting device

Introduction

Nowadays marketers face extensive challenges when

communicating with potential customers. The race for
getting the attention of a target audience is vast due to the
information overload of modern society. In an attempt
to stand out from other advertising messages, marketers
take the risk of getting trapped into the so-called vampire
effect of their own advertising campaigns: tools for
getting attention suck customer attention away from
core messages of the advertisement.
Despite the fact that a lot of studies have been done
concerning the effectiveness of advertising, little
attention has been drawn to this phenomenon. Thus,
this study aims to explore the risks of vampire effect in
advertising using celebrity endorsement and to provide
recommendations for preventing its occurrences.
Literature Review

Nowadays with the increasing amount of information

available worldwide, it gets rather hard for an advertiser
to create an effective marketing communication
campaign. The chase for reaching an audience is
becoming more complex, and getting people to sit down
16

to hear a message is challenging (Gobe, 2009, p. 243).

Furthermore, even when the message is delivered, it is
hard to ensure that the receiver understood it in the correct
manner. Besides, the targeted group could get overloaded
with marketing messages. Information overload is
generally dened as the state of an individual in which
not all communication inputs can be processes and
utilized, leading to breakdown (Quentin, 2004, p.196).
Cognitive psychologists proved that because a person
has a limited capacity to process information, he/she
creates a lter to prevent the information-processing
system from overloading (McLeod, 2008). Applying
this theory to marketing simply suggests that with being
exposed to too many advertising messages, a consumer
creates his/her own lter in order to deal with overload.
In the struggle for attention and getting their message
through to the target audience, marketers often use
so-called attention-getting devices (AGDs). AGD may
be dened as any element in a marketing communication
campaign designed to gain the attention of the audience
but which does not contain the core message of the
campaign. Humor, erotic stimuli and celebrities are
among the most commonly used AGDs.

CENTRAL EUROPEAN BUSINESS REVIEW

RESEARCH PAPERS

According to information overload theory, in times when

a person receives too much information, the captivation
of a piece of information occurs at the expense of another
piece of information (Salyer, 2013). This in turns means
that using AGDs should help in attention stimulation;
however, they could also distract customers from
receiving the necessary information about the product
and brand and, thus, create a vampire effect.
The vampire effect can generally be dened as an image,
object or person, which grabs the attention of the target
audience away from the product or the brand (or other
product/brand related messages) and, thus, prevents
customers from remembering the product or the brand.
A difculty in identifying these vampires in advertising
lies in the fact that any image used in an advertisement
has a risk of taking away the attention.
No general denition of the term vampire effect is
provided in academic literature; nevertheless, such
phenomena as vampire claims and vampire videos are
discussed which in turn help to understand the idea
behind it.
High risk of vampire effect is associated with using
celebrities in advertising. The probability of vampire
effect occurrence increases when there is no congruency
between the celebrity and a brand or a product. When
the celebrity endorsing the product does not have any
relation to the product he or she endorses, the vampire
effect is more likely to occur (Erdogan, 1999). Similarly,
a study of Speck, Schumann and Thompson found out
that related celebrities produced higher product recall
than unrelated celebrities (Erdogan, 1999). Nonetheless,
the difference was not statistically signicant, and thus
the outcome should be further veried.
Vampire effect in the context of celebrity endorsement
occurs when the celebrity overshadows the product
featured in the advertisement. A Cyber Media Research
study reveals a high percentage of vampire effect
occurrences: 80 percent of respondents in this study
could remember a celebrity but not the brand (Okorie,
Oyedepo and Akhidenor, 2012). It is also believed that
the probability of vampire effect occurrence increases if
an internationally unknown brand uses an international
celebrity (Erdogan and Baker, 1999). In such cases, it is
more likely that potential consumers will pay attention to
the celebrity and not to the brand.
An example of overshadowing and absence of clear
consistency with the product and celebrity advertising the
product may be Celine Dions endorsement by Chrysler
(Stein, 2003). After signing a three-year contract with
the singer for $14 million, the company soon realized
that effectiveness of celebrity endorsement turned out to

VOLUME 3, NUMBER 3 , SEPTEMBER 2014

be a disaster. Ms. Dion was featured in advertisements

for the 2004 Chrysler Pacica sport wagon, the Crossre
coupe and the Town & Country minivan. At that time,
the companys strategy was to move the brand towards
younger and more afuent customers. However, focus
groups revealed that the average age of audience to whom
Ms. Dion appealed was 52. The problem went so much
further that in the end the company had to reposition its
product and make the Pacica car targeting consumers
averaging 53 years old.
Therefore, vampire effect in advertising is an important
issue, since the occurrences of it may be quite hard to
predict. Thus, it needs to be studied in a greater detail.
Goal and Methodology
Based on a literature review, this study focused on
researching whether advertisements that use a celebrity
as an AGD were more likely to create a vampire effect
than those that do not use any AGD. However, the
differentiation between related and unrelated celebrities
as described by Speck, Schumann and Thompson
was used. Thus, the following hypothesis was formulated:
H1: Print advertisements that use unrelated celebrities
as AGD are more likely to create a vampire effect than
those that use related celebrities as AGD or those that use
no AGD at all:
a) Print advertisements that use unrelated celebrities as
AGD are more likely to create a vampire effect in
terms of taking attention away from the product.
b) Print advertisements that use unrelated celebrities as
AGD are more likely to create a vampire effect in
terms of taking attention away from the brand.
c) Print advertisements that use unrelated celebrities as
AGD are more likely to create a vampire effect in
terms of taking attention away from other product
relevant statements.
In order to test the hypothesis, three advertisements
were created (see Figure 1): one advertisement featuring
a celebrity unrelated to the product advertised (Angelina
Jolie used for promotion of a retro guitar store), one
advertisement featuring a celebrity related to the product
(Elvis Presley used for promotion of a retro guitar store)
and one advertisement not featuring any celebrity as an
AGD.
The study was performed in two phases. In the rst
stage, testing of printed ads was done with the help of
an eye-tracking device and included 12 participants
(university students). Participants had to take a look at
a series of ve advertisements, among which one of the
ads from Figure 1 was mixed.
17

CENTRAL EUROPEAN BUSINESS REVIEW

RESEARCH PAPERS

VOLUME 3, NUMBER 3 , SEPTEMBER 2014

Figure 1 Tested advertisements

Figure 1a-c
1a-c (clockwise
from top
left):top left):
Figure
(clockwise
from
1a: unrelated celebrity advertisement;
related celebrity advertisement;
1a:1b:unrelated
celebrity advertisement;
1c: advertisement without a celebrity.

1b: related celebrity advertisement;

1c: advertisement without a celebrity.

Specically, four participants were exposed to an ad

with a non-related AGD, four participants to an ad with
a related AGD, and four participants to an ad without
any AGD. Each ad was shown to the participants for
a period of ve seconds. Afterwards, participants were
asked to ll in a survey. After collecting information,
the following outputs from the eye-tracking device
were generated: heat maps, scan paths, as well as
clustering of gaze. Based on this, the researchers
could see typical ways each advertisement was looked
at as well as attention distribution on each of them.
Additionally, the results of this stage helped to adjust
the questions for the survey used more broadly in the
next stage.
The second phase was an experiment with 60 students
(university students as well). During this stage, participants were divided into three groups of 20 people and
exposed to the same mixture of advertisements in the
same manner as in the rst stage. Thus, the rst group
saw the advertisement with an unrelated celebrity as
18

AGD, the second advertisement with related celebrity

as AGD and, nally, the third group advertisement
with no celebrity as AGD. It is necessary to notice
that the survey stage also had some qualitative inputs:
respondents had to answer open-ended questions. Thus,
the research approach was a combination of exploratory
and conclusive research. Afterwards, students were
asked to ll in a survey. Subsequently, the results were
inserted into SPSS for conducting statistical analysis. The
researchers used the Kruskal-Wallis test for identifying
whether there were signicant differences between the
three sample groups in recall levels of different parts of
the tested advertisements.
Findings

As discussed previously, a general denition of the

vampire effect suggests that an AGD eats the attention
of the target group away from the core messages of
the advertisement brand, product and other messages
related to the product.

CENTRAL EUROPEAN BUSINESS REVIEW

RESEARCH PAPERS

The researchers analyzed the survey results using

a non-parametric test of comparing means. This was done
due to the fact that data was not normally distributed and
since the number of independent groups to compare was
more than two, the Kruskal-Wallis test was chosen for
analysis. The necessary assumptions of homogeneity of
variances and similar distributions were met for each
analysis performed.
Survey analysis showed that in terms of levels of
remembering of the ad image, ads with celebrity
endorsement showed better results than the ad picturing
a guitar (see Figure 2). Further statistical testing using
the Kruskal-Wallis test showed that the difference
between the groups was signicant (Chi-square = 7,549;
p = 0,023 < 0,05). Specic comparison testing revealed
that a statistically signicant difference in remembering
AGD was between group 1 and group 3 (Chi-square =
4,789; p = 0,029 < 0,05); and between groups 2 and 3
(Chi-square = 6,240; p = 0,012 < 0,05). The control
group had a signicantly lower image recall level (Mean
ranks: G1= 33,5; G2= 35,0; G3= 23,0). These results
were also consistent with the results of the eye-tracking
experiment: participants exposed to advertisements with
celebrity endorsement drew more attention to the faces
of celebrities, whereas participants from the control
group drew less attention to the picture of a guitar.

VOLUME 3, NUMBER 3 , SEPTEMBER 2014

38,0; G3= 32,0). In other words, even though participants

could much better remember a celebrity (Angelina Jolie)
in the ad they saw, they did not remember which product
she endorsed. The null hypothesis of this part (a) was
therefore rejected. It was thus concluded that an unrelated
celebrity ad showed the highest risk of vampire effect.
In relation to part b) of the hypothesis drawing
attention away from the brand no statistical proof
was found. Participants generally showed a very low
level of brand recall across three groups as can be seen
in Figure 3. Groups two and three had the same recall
levels of 20 percent; however, unrelated celebrity ad
(group 1) yielded no recall of brand information at all.
However, the Kruskal-Wallis test revealed that there was
no statistically signicant difference between the three
groups (Chi-square = 4.538; p = 0.103 > 0.05).
Figure 3: Comparison of ad image and brand recall levels

Figure 2 Comparison of the ad image and product recall

levels

Source: authors

However, the levels of product recall had reverse results

(see Figure 2). There was a signicant difference between
the three groups in terms of product recall (Chi-square =
12,701; p = 0,002 < 0,05). Specically, there was
a difference between groups 1 and 2 (Chi-square =
12.584; p = 0.00 < 0.05) and groups 1 and 3 (Chi-square =
5.991; p = 0.014 < 0.05). Thus, ads with unrelated
celebrity endorsement (G1) showed signicantly lower
levels of product recall (Mean ranks: G1= 21,5; G2=

The null hypothesis of this part was hence not rejected.

There was no signicant difference between the three
groups in terms of brand recall. These results are also
consistent with the output from the eye-tracking experiment: the eye-tracking device spotted almost no direct
look at the brand logo. Participants looked at the logo
mostly indirectly, without actively seeing the name of the
company. This lack of attention towards the brand could
be due to the wrong placing of the logo in the advertisement or the limited amount of time given to participants.
Further discussion of reasons for not having a direct
attention drawn to the logo of the company, however,
goes beyond the scope of this study.
Speaking about part c) of the hypothesis, the related
AGD ad had a signicantly higher level of other product
related statements recall than the other two advertisements. Survey results showed the best recall levels in the
group exposed to related AGD ad (70%) as can be seen
in Figure 4. The lowest recall levels were observed in the
groups with respondents seeing an unrelated celebrity as
AGD advertisement only 5 percent could remember
other statements. Performing the Kruskal-Wallis test
19

CENTRAL EUROPEAN BUSINESS REVIEW

VOLUME 3, NUMBER 3 , SEPTEMBER 2014

RESEARCH PAPERS

Table 1 Measures of association levels between celebrity and product recall for test group 2
Chi-Square Tests and Symmetric Measures

Pearson Chi-Square

Value

df

Asymp. Sig.
(2-sided)

5.089a

.024

Exact Sig.
(2-sided)

Fishers Exact Test

Cramers V
N of Valid Cases

Approx.
Sig.

.050

0.504

0.024

20

Source: authors

showed that the difference was statistically signicant

(Chi-square = 14,359; p = 0,001 < 0,05). Specically,
the difference laid between groups 1 and 2 (Chi-square =
12,769; p = 0,000 < 0,05) and groups 2 and 3 (Chi-square =
6,240; p = 0,012 < 0,05). Related AGD advertisement
had the highest recall levels (Mean ranks: G1= 23,0;
G2= 40,15; G3= 28,35).
Figure 4: Comparison of ad image and other statements
recall levels

Source: Authors

The null hypothesis can thus be rejected. The unrelated

AGD ad had still lower levels of other product related
statements recall, which, together with AGD recall
levels, suggests that this ad showed the highest risk of
a vampire effect as compared to the related AGD ad or to
the ad without any AGD. Compared to the results of the
eye-tracking experiment, this conclusion is consistent as
well: the gaze scans of the unrelated AGD ad showed
almost no coverage of product/brand related statements,
whereas gaze scans of the related AGD and non AGD ad
partially covered these statements.
Noteworthy, however, was also the fact that even though
the gaze scan coverage of the control ad was much broader
20

and covered other product/brand related statements

much better, the recall levels of the participants were
signicantly lower as compared to the recall levels of
participants who saw a related celebrity as AGD ad.
Yet, there was no statistically signicant difference in
other statement recalls between the control ad and the
unrelated celebrity as AGD ad. More detailed analysis is,
therefore, necessary to explain this phenomenon.
Therefore, since two out of three subparts of the
hypothesis were not rejected, the research concluded that
overall, the null hypothesis of the study was rejected.
Therefore, the study found that there is a signicantly
higher risk of creating a vampire effect when using an
unrelated celebrity in print advertising than when using
a related celebrity or no celebrity at all. Furthermore,
related celebrity endorsement could be an effective tool
in printed advertising. However, the level of relatedness
of an attention-getting device can rarely be measured
precisely since perceiving a celebrity related to a product/
brand advertised is subjective.
Furthermore, using no AGD has a risk of being ignored by
the target market. As revealed by testing, the perception
levels of advertisements being able to effectively grab
attention of the audience, the advertisement with no
celebrity as AGD (G3) had the worst results: only
5 percent of respondents believed the ad effectively
grabbed their attention (G1 = 45%; G2 = 35%). Statistically, the difference was signicant (Chi-square =
9,486; p = 0,009). The difference laid between groups
2 and 3 (Chi-square = 8.142; p = 0.004 < 0.05) and groups
1 and 3 (Chi-square = 6.080; p = 0.014). The mean ranks
of the Kruskal-Wallis test were the following: G1= 35,1;
G2= 35,6; G3= 20,8. Indeed, the control group was much
less likely to perceive the advertisement they saw to be
catchy, whereas the groups exposed to advertisement
with celebrities as AGD had the opposite result.

CENTRAL EUROPEAN BUSINESS REVIEW

RESEARCH PAPERS

Additionally, the hypothesis that the risk of vampire

effect is higher for unrelated celebrity advertisement is
also supported by the perception of advertisement consistency. The analysis of the survey results revealed that
the advertisement that used unrelated celebrity as AGD
had the worst perception of consistency 20 percent of
respondents thought the ad was consistent as compared
to 55 percent in group 2 and 25 percent in group 3.
Participants in the group had the weakest belief that all
messages in the advertisement made sense all together.
Statistical analysis of signicance conrmed there
was a difference (Chi-square = 5,972, p = 0,05 = 0,05)
between the three groups. Mean rank distributions were
the following: G1= 25,53; G2= 38,08; G3= 27,9. Yet, the
difference was only between groups 1 and 2 (Chi-square =
5.773; p = 0.017 < 0.05). There was, however, no statistically signicant difference between groups 1 and 3
(Chi-square = 0.109; p = 0.741 > 0.05), as well as
between groups 2 and 3 (Chi-square = 3.070; p = 0.080
> 0.05). Even though no deeper analysis of such perception was done, it could be assumed that unrelated celebrity created an imbalance in communication effort of the
company and, thus, could eventually lead to a vampire
effect. Coming back to previous studies, Erdogan (1999)
had already theorized that when a celebrity endorser did
not have an associated relation to the product, a vampire
effect tended to occur. The fact that the control advertisement was also perceived as quite inconsistent, since only
25 percent of participants believed it was, could be due
to the low levels of the ad recall.
Furthermore, as was discussed previously, Speck, Schumann and Thompson (Erdogan, 1999) found out that
related celebrities produced higher product recall than
unrelated, but did not have statistical proof for that. This
study shed more light on the discussion. The statistical
analysis of survey results showed positive statistical
association between the levels of the celebrity recall and
the level of product recall in the group exposed to the
advertisement with related celebrity (Table 1). Thus, the
theoretical belief that in cases of using celebrity endorsements, if done correctly by achieving the degree of consistency between the product and the celebrity endorsing
it, movement of meaning may occur. In other words, if
consumers see a logical link between an endorser and
a product, they are more likely to remember the product.
Conclusion and Managerial Implications

Marketers should be able to nd a balance between

gaining the attention of their target audience and
delivering product and brand related message. Gaining
attention is the rst step only; it is a means to an end rather

VOLUME 3, NUMBER 3 , SEPTEMBER 2014

than an end itself. As warned by previous researchers,

this study showed once again that the race for creativity
and attention getting could easily become a false priority.
If the level of association between the AGD and the
product or brand is not sufcient enough in the minds of
the target audience, no value will be generated from such
advertisement. The fact that the audience remembers
the advertisement does not automatically mean they
received the core message about the product or a brand.
Using unrelated celebrities as AGD has a higher risk of
creating a vampire effect. Nevertheless, even though an
advertisement without any special AGD could be a safer
option, it has a higher risk of being simply ignored by
target groups. Therefore, in order to pass through the
lter of consumers and get their attention, marketers
might still need to use special AGDs. Consequently,
when choosing to use celebrities as AGD, marketers
should use related celebrities since, as discussed earlier
by Speck, Schumann and Thompson (Erdogan, 1999)
and proven by this study as well, they produce higher
product recall. Nevertheless, this relation should be
pre-tested prior to launching an advertising campaign.
However, it should be noted that levels of relatedness
between the celebrity and the target audience is hard
to test, since this is a subjective and individual matter.
Thus, in this regard, the study faces its limitations.

References
Erdogan, B.Z. (1999). Celebrity Endorsement: A Literature
Review. Journal of Marketing Management. 15 (1):
291-314.
Erdogan, B. Z., Baker, M. J. (1999). Celebrity Endorsement:
Advertising Agency Managers Perspective. The CyberJournal of Sport Marketing, 3 (3).
Gobe, M. (2009). Emotional Advertising: Expressing Real
Emotions Online, in Emotional Branding: The new
paradigm for connecting brands to people. New York:
Allworth Press, 223-244.
Heath, R. (2012). Problems with Getting Attention, in
Seducing the Subconscious: The Psychology of Emotional
Inuence in Advertising. Hoboken: Wiley, 86-98.
McLeod, S. (2008). Selective Attention. Simply Psychology.
(accessed January 20, 2014), [available at http://www.
simplypsychology.org/attention-models.html]
Okorie, N., Oyedepo, T., Akhidenor, G. (2012). The
Dysfunctional and Functional Effect of Celebrity
Endorsement on Brand Patronage. Online Journal of
Communication and Media Technologies,
2 (2): 148-149.

21

CENTRAL EUROPEAN BUSINESS REVIEW

RESEARCH PAPERS

Quentin, J. (2004). Information Overload and the Message

Dynamics of Online Interaction Spaces: A Theoretical
Model and Empirical Exploration. Information Systems
Research, 15 (2): 194-210.
Salyer, G. (2013). Information Overload: The effects of
advertising avoidance on brand awareness in an online
environment. (accessed January 20, 2014), [available at
http://web02.gonzaga.edu/comltheses/proquestftp/Salyer_
gonzaga_0736M_10263.pdf ]
Stein, J. (2003). Inside Chrysler and Celine Dion Advertising
Disaster. Advertising Age. November 24.

This study was realized in cooperation with INCOMA GfK

research agency and it was supported from the resources

for long-term conceptual research development of the
University of Economics, Prague (IP300040).

22

VOLUME 3, NUMBER 3 , SEPTEMBER 2014

Authors
Tetyana Kuvita
M.A. in International Business
University of Applied Sciences, Mainz
Koenigsgartenstrasse 1, 55583,
Bad Muenster-Ebernburg, Germany
tetyana.kuvita@gmail.com
Assoc. Prof. Miroslav Karlek, Ph.D
Marketing Department
University of Economics, Prague
Nm. W. Churchilla 4